Table 2.

Adjusted Logistic Regressions of the Association Between Health Information Technology Components and Preventable Emergency Department Visits for ADRD Patients.

Variable	OR	95% CI	p-value
Health Information Technology Patient Engagement Functionalities
Low Engagement Functionalities (0–7)	Ref
High Engagement Functionalities (8–9)	0.90	0.85–0.95	<0.001
Individual HIT-Patient Engagement Function
Patient can electronically send care/referral summaries to a third party	0.89	0.83–0.96	0.002
Patient can request an amendment to change/update their health/medical record	0.85	0.79–0.92	<0.001
Patient can request refills for prescriptions online	0.89	0.84–0.95	<0.001
Patient can submit self-generated data	0.93	0.87–0.99	0.031
Routinely Integrate Electronic Clinical Information from Outside Sources
No	Ref
Yes	1.05	0.99–1.12	0.121
Routinely Have Clinical Information Available Electronically from Outside Providers
No	Ref
Yes	0.98	0.92–1.04	0.550
Don’t Know	1.03	0.85–1.25	0.743
Often Use Electronic Patient Health Information from Outside Providers
No	Ref
Yes	0.88	0.82–0.95	<0.001
Provide Electronic Notification to the Patient's Primary Care Physician
Only Inside System	Ref
Inside and Outside System	0.91	0.84–0.98	0.013
Don't Notify/Don't Know	1.03	0.95–1.10	0.501

Notes. Abbreviations: OR = adjusted odds ratio, CI = confidence interval. These are results from four separate regressions. Each estimation function adjusted for patients’ race, gender, age, insurance, zip code income quartile, Elixhauser comorbidities, county urban/rural status, county percent African American, county HPSA status, county MHPSA status, hospital number of beds, hospital ownership, last quarter index, and state fixed effects. Sensitivity tests were performed based on different cutoffs for preventable ED visits (40% and 60%). Sample size: 108,828. Full results are available upon request.